OpenStitch.
A free, open-source, cross-platform embroidery digitization app. Built to replace the legacy paid tools that lock the craft behind a $799 paywall and a Windows-only install.
Embroidery digitization is stuck in 2003.
Hatch Embroidery 3 ships at $799. It runs only on Windows. Its file format is opaque even to itself. It can read its own old designs but can not always write them back without breakage. ML auto-digitization is non-existent. Tutorials live behind paywalls. The tooling assumes you already know what you are doing.
If you are a working embroiderer, you live with this. If you are starting out, the cost is the first reason to quit.
A modern free alternative.
Built ground-up in Rust + React. Cross-platform. Open source. Two superpowers Hatch lacks: full .EMB round-trip, and a photo-realistic ML auto-digitizer.
46 file formats
Read and write every major format including PES, DST, JEF, EXP, VP3 via the PyEmbroidery sidecar. 40 round-trip regression tests passing.
Full .EMB round-trip
Read and write Wilcom Hatch native designs. All 7 streams parsed and re-emitted. Goal: edit a Hatch design in OpenStitch, save, reopen in Hatch, no loss.
ML auto-digitizer
Drop a photo, get a stitch file. Classical OpenCV path for logos plus a neural DINOv2 + UNet path for photo-realism. Runs fully offline.
Offline AI tutor
Phi-3.5-mini bundled and lazy-downloaded (~2.2GB). Asks you questions, answers yours, walks you through your first design without a sales pitch.
7 digitizing tools
Select, Running, Manual, Satin column, Tatami fill, Motif fill, Appliqué. All shipped, all keyboard-driven, all undo / redo aware.
Stitch simulator
Watch your design stitch out before you load thread. Playback bar, density preview, color callouts, smart auto-zoom on color change.
A serious modern stack.
Tauri 2
Rust backend. Smaller, faster, safer than Electron.
React 19 + PixiJS 8
GPU-accelerated canvas, 60fps stitch rendering.
PyEmbroidery
Bundled Python 3.11 sidecar handles legacy formats.
llama.cpp + ONNX
Local LLM tutor + ML auto-digitizer. Zero API cost.
A craft tool should not gatekeep the craft.
OpenStitch is licensed under AGPL-3.0. The full source code lives on GitHub. Anyone can read it, audit it, fork it, run it, or contribute back. There is no enterprise tier, no premium unlock, no cloud dependency. The only paid dependency in the entire pipeline is a one-time ~$60 cloud GPU rental during ML training, which we cover.
We are building OpenStitch because the alternative is to keep paying for software that has barely changed in fifteen years. Free embroidery software exists, but none of it covers the full pipeline. We are bringing the full pipeline to free.
Where we are right now.
Phase 0 to 3
Scaffold, file I/O, .EMB round-trip code, canvas, all 7 manual digitizing tools.
Phase 9A (Animation)
Export-as-animation pipeline: PixiJS render to MP4 / WebM / GIF with social presets.
Phase 4, 4B, 5
Auto-digitizer (classical + neural), offline tutor inference body, real-time recording.
Get notified when it ships.
No spam. We will email you once when the first public build is downloadable, and once when the auto-digitizer hits acceptable quality. That is it.
Notify me →